Post-processing

Finally, once we get the summary, as it is uncased, we make use of a truecaser model to case it correctly.

Fig. 5 Truecasing of the generated summary.

Currently, we use the truecase module for Python, which implements a statistical model inspired by the paper tRuEcasIng by Lucian Vlad Lita et al.

At this point, the summary is ready to be delivered to the user.